Search                        Top                                  Index
HELP ARRAYS                                      Jon Cunningham Feb 1982
                                               Updated A.Sloman May 1990

Information on POP-11 arrays

CONTENTS - (Use <ENTER> g to access desired sections)

 -- Introduction
 -- The structure of Arrays
 -- Creating arrays using NEWARRAY
 -- Initialising an array with a procedure
 -- Arrays of characters (byte arrays)
 -- More complex cases: NEWANYARRAY etc
 -- Two arrays may share data
 -- Arrays in POP-11 are procedures
 -- Arrays are data-structures
 -- Associated procedures
 -- Non-numerical subscripts
 -- Related documentation


-- Introduction -------------------------------------------------------

Arrays are multi-dimensional objects that can be used to store data
accessed or updated by means of numerical subscripts. An N-dimensional
array requires N integers as subscripts to access or update its
components. Two dimensional arrays are often used to represent visual
images.

Ordinary arrays can contain any POP-11 objects including arrays
themselves. Facilities are provided (See HELP * NEWANYARRAY) for
creating an array restricted to 'packed' numerical data or bit patterns.
For instance an array may be restricted to N bit integers, where 1 <= N
<= 32. Such arrays use vector types that may be defined by the user. For
details see REF *DEFSTRUCT


-- The structure of Arrays --------------------------------------------

An array in POP-11 is a structure composed of:

    1. a vector-type object containing the data
    2. a 'boundslist' specifying the dimensions of the array and
       the bounds for each dimension
    3. an accessing/updating procedure that computes which component
       of the vector corresponds to the numerical subscripts given when
       the array is used.

For this reason an array will take up more space than a vector with the
same components. However arrays are very convenient when representing
two, three or N-dimensional structures.


-- Creating arrays using NEWARRAY -------------------------------------

The simplest procedure for creating arrays is -newarrray-. Its use takes
various formats, the most common being

    newarray(<boundslist>, <initial value>) -> <array>

For example,

    newarray([1 10 1 10], 0) -> aa;

will create and assign to aa, a 10 by 10 two dimensional array each of
whose elements is 0 initially. Arrays are printed out including the
boundslist:

    aa =>
    ** <array [1 10 1 10]>

The elements are accessed by means of two integers in the ranges of the
boundslist.

    aa(3,9) =>
    ** 0

If the subscripts are out of bounds a mishap results.

    aa(3,29) =>
    ;;; MISHAP - INVALID ARRAY SUBSCRIPT
    ;;; INVOLVING:  29
    ;;; DOING    :  compile nextitem popval compile

The bounds may be negative, as in: newarray([-5 5 -9 90], []).

The variable -poparray_by_row- controls the way elements are stored. If
it is -true- (the default) then the first parameter of arrays determine
adjacency in underlying vector. If -false-, then the last parameter.

If the second argument of -newarray- is missing it defaults to the word
"undef". (See HELP * UNDEF). The second argument may be the empty list
[], but not any other list as it will be confused with the boundslist.
If it is a procedure then the procedure will be used to compute initial
components of the array, as follows.


-- Initialising an array with a procedure -----------------------------

    newarray(<boundslist>, <procedure>) -> aa;

where <boundslist> has 2N elements and <procedure> takes N integers and
returns one result will create and assign to aa an array whose elements
are initialised to be the result of applying the <procedure> to the
subscripts. (For a simple example see HELP * NEWARRAY.) So the value of
aa(i1, i2, ... in) will be the same is the result of applying the
procedure to (i1, i2, ... in). However, the values can be changed, since
every array has an 'updater' (See HELP * UPDATER). Thus:

    <item> -> aa(i1, i2, ... in)

will change the value stored in the location accessed by those
subscripts.


-- Arrays of characters (byte arrays) ---------------------------------

newsarray(<list:B>) -> <array>
newsarray(<list:B>, <character:CHAR>) -> <array>
    Constructs a character array with bounds B and cells initialised to
    CHAR (see LIB * NEWSARRAY)


-- More complex cases: NEWANYARRAY etc --------------------------------

More complex options, including building arrays out of user-defined
vector-types are provided by -newanyarray-. Sparse arrays may be created
by -newsparse-, and -newanysparse- provides more flexible facilities.
For details and examples see the following help files:

    HELP * NEWARRAY * NEWANYARRAY
        Creating arrays

    HELP * NEWSPARSE * NEWANYSPARSE
        Creating "sparse" arrays with properties


-- Two arrays may share data ------------------------------------------

It is possible for two arrays to share the same vector or a part of a
vector, whilst viewing it in different ways, for instance using
different boundslists. Thus a 12 by 12 array and a 3 by 6 by 8 array may
have the same vector accessed in the one case by two integers and in the
other by three. For details see HELP * NEWANYARRAY.


-- Arrays in POP-11 are procedures ------------------------------------

As the above implies, POP-11 allows an N-dimensional array to be used in
any context in which a procedure of N integer arguments and one result
may be used, and therefore procedures like -pdprops-, -partapply-,
-updater-, -pdnargs- may be applied to them. Closures may be formed as
with procedures. E.g. if aa is a 3-D array then aa(%5%) will be a 2-D
array obtained by 'freezing' the third subscript at the value 5. (See
HELP * CLOSURES)

-isprocedure- will return -true- if applied to an array. -isarray- can
be used to distinguish arrays from procedures (though it returns the
array itself, not -true-, if the argument is an array).

If something is assigned to the -pdprops- of an array it will alter the
way the array is printed. E.g.

    "demo_array" -> pdprops(aa);
    aa =>
    ** <array demo_array [1 10 1 10]>

Since the datakey of an array is the procedure key, arrays as such do
not have their own -class_print-. (See HELP * CLASSES)


-- Arrays are data-structures -----------------------------------------

Many of the procedures that apply to POP-11 data-structures can be
applied to arrays. See, for example *DATALENGTH  *EXPLODE *DATALIST
*APPDATA and *COPY


-- Associated procedures ----------------------------------------------

The procedure -boundslist- applied to an array will return the list of
bounds given as first argument to -newarray- or -newanyarray-.

The procedure -arrayvector- returns the vector of the array (see HELP
*ARRAYVECTOR.)

The procedure -arrayvector_bounds- returns the bounds of the used
portion of this array (normally 1 and the length of the array.)


-- Non-numerical subscripts -------------------------------------------

Arrays store association between sets of numbers and arbitrary values.
POP-11 also provides mechanisms for storing associations between
arbitrary objects:

See HELP *NEWASSOC, *NEWPROPERTY, *NEWANYPROPERTY, *NEWANYSPARSE
*NEWMAPPING, and REF *ASSOC.


-- Related documentation ----------------------------------------------

For technical details on arrays, see REF * ARRAYS

For a library to print arrays as tables as numbers or as blocks of
'grey-level' characters, see TEACH * SHOWARRAY. For a library to read
blocks of characters (like LIB * GREYPIC) into arrays, see HELP
*READARRAY.

For more general information on POP-11 data-structures see:

    HELP *DATASTRUCTURES *VECTORS *CLASSES *KEYS

    REF *DATA *KEYS *VECTORS *DEFSTRUCT *ARRAYSCAN

For generalised mappings from arbitrary objects to arbitrary objects see
HELP *PROPERTIES and REF *PROPS



--- C.all/help/arrays
--- Copyright University of Sussex 1992. All rights reserved. ----------